correction in active cases data of covid-19 for the us ......2020/09/02 · zhonghua liu xing bing...

1

Correction in Active Cases Data of COVID-19 forthe US States by Analytical study

Ravi Solanki1,Anubhav Varshney2,Raveesh Gourishetty3 Saniya Minase4,Namitha Sivadas5,Ashutosh Mahajan5˚1Visvesvaraya National Institute of Technology, Nagpur-440 010, India2 Swami Vivekanand College of Engineering, Indore-452 020, India3 Indian Institute of Technology Bombay, Mumbai-400076, India

4 Birla Institute of Technology and Science Pilani, Goa-403 726, India5Vellore Institute of Technology, Vellore-632 014, India

˚E-mail:[email protected]

Abstract—The reported recovered cases for many US statesare low which could be due to difficulties in keeping track ofrecoveries and resulted in higher numbers for the reported activecases than the actual numbers on the ground. These incorrectnumbers can lead to misleading inferences. In this work, basedon the typical range of recovery rate of COVID-19, we estimatethe active data from the total cases and death cases and bringout a correction for the data for all the US states reported onworldometer.

Keywords— COVID-19, Mathematical Model, epidemiology

BACKGROUNDThe availability of accurate data of an epidemic is important as

the data provides key insights on the disease spread and enable theauthorities to take a decision on control measures. Worldometer isone of the very popular sources of the global COVID-19 data andit is also trusted and used by many government bodies and agencies[1]. The available data for the COVID-19 cases can be used for theprediction and analysis of hospitalization and meeting the demandsof health care facilities and setting up the critical care systems forthe patients [2] [3]. The active cases represent the number of infectedpeople whether symptomatic or asymptomatic detected through self-reporting or testing. This number is important for Public Healthauthorities to estimate the current status of the disease spread andcan be calculated by subtracting death and recovered cases from thetotal confirmed cases.

METHODA compartmental predictive mathematical model SIPHERD for

COVID-19 disease dynamics is described in [4], [5] where recoveryrate is a model parameter and is fixed by optimizing the model withthe actual data. The data for the total, death, and active cases for168 days from March 4, 2020, is taken from Worldometer [1] andfound close to total cases and death data from [6]. After running theSIPHERD model for the USA, as reported in [4], the recovery rateof the active category was found to be 0.015 (corresponding to 66days of mean recovery time) which is very low compared to othercountries like Germany and India [4] [5] where it is found 0.065. Thelow recovery rate in the USA may be attributed to either incorrectreporting of the Active cases [7] or the testing of serious casesonly and longer recovery time in hospitals compared to quarantinedwith mild symptoms. Secondly, keeping the record of recoveries isdifficult because some of the infected people are asked to quarantinewhile only critical patients are hospitalized. Sometimes the reportingof those recoveries is not accurate or incomplete. This has led toinconsistent data for active cases.

The number of mild cases is reported to be 81% in a Chinesestudy [8]. Covid-19 data reported from 49 states, the District ofColumbia, and three U.S. territories to CDC from February 12–March16 shows that 20.7 reported cases were severe and were hospitalized[9]. COVID-NET regions show this number to be 21.4 % till April4 [10], [11] and IHME data March 5- April 4 shows that to be 20.3% [12] [13]. According to WHO recovery time for mild cases is of2 weeks and for severe cases that recovery time of 3-6 weeks. [14].Considering 80% mild cases, the recovery rate cannot be as low asappears from the data for the states listed in column 2 in Table 1.

The correct estimation of the active cases can be done bysubtracting the death and recovery cases, with appropriate recoveryrate, from the total cases. The woldometer data for total cases anddeath cases are assumed to be true as the test positive results andcount of diseased cases are done more stringently as compared torecovery counting. The active cases can be obtained by solving thefollowing differential equation,

dHptqdt

“ dT ptqdt

´ dDptqdt

´ σHpt´ tRq (1)

where, H , T , and D are the active, total, and death cases. Therecovery from the infected category is defined by the two parameters:delay in recovery tR and the recovery rate σ. As these valuesare dependent upon the immune system of the community and thehospital facilities, it should not vary much within the USA. We havetaken tR as 10 days and σ as 0.048 (21 days of mean recovery timeconsidering both mild and severe cases).

RESULTS AND DISCUSSION

The above delay differential equation is solved for all the states ofthe USA and we found three groups among the states according tothe accuracy of the data. The active cases reported on worldometer[1] for a few states show excellent agreement with our estimation ofactive cases. One example state for this group is Texas as seen in Fig.1(a). There are few states in the second group which are largely notmatching with analytical estimation indicating that reported activedata is inaccurate. These states are listed in column 2 in Table 1 andone representative state is Virginia as seen in Fig. 1(b), where thecurrently active cases are reported to be 97560 which should havebeen just around 21664 according to our calculation. Interestingly,in the last group, there are some states for which the reported activecases do not follow the estimated active cases for the initial days,however, the reported data of the active cases take a sudden fall andfollow our estimation as seen for Indiana in Fig. 1(c). This clearlyindicates that the active cases data corrections is done after some daysand also validates our analytical approach. In Fig. 1, the reported totaland active cases with the estimated active cases for one of the states

. CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprintthis version posted September 3, 2020. ; https://doi.org/10.1101/2020.09.02.20186833doi: medRxiv preprint

NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice.

https://doi.org/10.1101/2020.09.02.20186833http://creativecommons.org/licenses/by/4.0/

2

(a) (b) (c)

Fig. 1: (a) Texas representing states in which active data is corrected reported. (b) Virginia represents the second group wheredata is largely incorrect and (c) Indiana from the third group where data has been corrected recently and closely match withanalytical estimation.

TABLE I: The US States Active Cases Data Status

Correct Data Incorrect Data Recently CorrectedTexas Georgia NebraskaTennessee New York IndianaMichigan Arizona IllinoisOhio Maryland MassachusettsAlaska Colorado PennsylvaniaNorth Dakota Delaware Dist of ColombiaNew Hampshire Idaho VermontSouth Dakota New Mexico LouisianaWest Virginia Minnesota AlabamaWyoming Rhode IslandMaine MissouriUtah WashingtonNorth Carolina KentuckyMontana ConnecticutWisconsin VirginiaIowa CaliforniaHawaii New JerseyArkansas FloridaMississippi South CarolinaKansas Oregon

Oklahoma

in all the three groups are shown, the figures for remaining states aregiven in supplementary material.

The recovery rate in individual states may vary to an extent of˘10% depending on the variation in the number of tests performed,fraction of mild and severe cases. However, we have taken a uniformvalue of recovery rate as 0.048 for all the US states.

CONCLUSIONSThe reported active cases for a few states is consistent with the

total detected cases and death cases for a recovery rate parametervalue 0.048. For a few states, the data has been corrected recently.However, for 21 states, the active cases data is still largely incorrect.We generate the corrected active case data for all states, report in thesupplementary material, and also keep it available on Github.

A. Data AvailabilityThe data for the corrected active case for all the US States can be

downloaded with the github link below.https://github.com/ravisolankigithub/covid-activecases-usa.git

REFERENCES[1] https://www.worldometers.info/coronavirus.

[2] S. M. Moghadas, A. Shoukat, M. C. Fitzpatrick, C. R. Wells, P. Sah,A. Pandey, J. D. Sachs, Z. Wang, L. A. Meyers, B. H. Singer et al.,“Projecting hospital utilization during the covid-19 outbreaks in theunited states,” Proceedings of the National Academy of Sciences, vol.117, no. 16, pp. 9122–9126, 2020.

[3] J. J. Cavallo, D. A. Donoho, and H. P. Forman, “Hospitalcapacity and operations in the coronavirus disease 2019 (covid-19)pandemic—planning for the nth patient,” in JAMA Health Forum, vol. 1,no. 3. American Medical Association, 2020, pp. e200 345–e200 345.

[4] A. Mahajan, R. Solanki, and N. Sivadas, “Estimation of undetectedsymptomatic and asymptomatic cases of covid-19 infection andprediction of its spread in usa,” medRxiv, 2020. [Online].Available: https://www.medrxiv.org/content/early/2020/06/23/2020.06.21.20136580

[5] A. Mahajan, N. A. Sivadas, and R. Solanki, “An epidemic model sipherdand its application for prediction of the spread of covid-19 infection inindia,” Chaos, Solitons & Fractals, p. 110156, 2020.

[6] https://covidtracking.com/data.[7] [Online]. Available: https://www.usnews.com/news/health-news/

articles/2020-04-02/why-are-us-coronavirus-recovery-numbers-so-low[8] C. P. E. R. E. Novel et al., “The epidemiological characteristics of

an outbreak of 2019 novel coronavirus diseases (covid-19) in china,”Zhonghua liu xing bing xue za zhi= Zhonghua liuxingbingxue zazhi,vol. 41, no. 2, p. 145, 2020.

[9] C. COVID and R. Team, “Severe outcomes among patients withcoronavirus disease 2019 (covid-19)—united states, february 12–march16, 2020,” MMWR Morb Mortal Wkly Rep, vol. 69, no. 12, pp. 343–346, 2020.

[10] [Online]. Available: https://gis.cdc.gov/grasp/covidnet/COVID19 3.html

[11] S. Garg, “Hospitalization rates and characteristics of patientshospitalized with laboratory-confirmed coronavirus disease2019—covid-net, 14 states, march 1–30, 2020,” MMWR. Morbidityand mortality weekly report, vol. 69, 2020.

[12] https://covid19.healthdata.org/united-states-of-america.[13] I. COVID, C. J. Murray et al., “Forecasting the impact of the first wave

of the covid-19 pandemic on hospital demand and deaths for the usaand european economic area countries,” medRxiv, 2020.

[14] [Online]. Available: https://www.who.int/dg/speeches/detail/who-director-general-s-opening-remarks-at-the-media-briefing-on-covid-19---24-february-2020



https://www.medrxiv.org/content/early/2020/06/23/2020.06.21.20136580https://www.medrxiv.org/content/early/2020/06/23/2020.06.21.20136580https://www.usnews.com/news/health-news/articles/2020-04-02/why-are-us-coronavirus-recovery-numbers-so-lowhttps://www.usnews.com/news/health-news/articles/2020-04-02/why-are-us-coronavirus-recovery-numbers-so-lowhttps://gis.cdc.gov/grasp/covidnet/COVID19_3.htmlhttps://gis.cdc.gov/grasp/covidnet/COVID19_3.htmlhttps://www.who.int/dg/speeches/detail/who-director-general-s-opening-remarks-at-the-media-briefing-on-covid-19---24-february-2020https://www.who.int/dg/speeches/detail/who-director-general-s-opening-remarks-at-the-media-briefing-on-covid-19---24-february-2020https://www.who.int/dg/speeches/detail/who-director-general-s-opening-remarks-at-the-media-briefing-on-covid-19---24-february-2020https://doi.org/10.1101/2020.09.02.20186833http://creativecommons.org/licenses/by/4.0/

3

SUPPLEMENTARY MATERIAL

a b c

e. f. f.

e. f. f.

Fig. S1: Reported total cases, active Cases data and analytically estimated active cases for various US states




4

a b c

e. f. f.

e. f. f.





5

a b c

e. f. f.

e. f. f.





6

a b c

e. f. f.

e. f. f.





7

a b c

e. f. f.

e. f. f.





8

a b c

e. f. f.





Data AvailabilityReferences

correction in active cases data of covid-19 for the us ......2020/09/02 · zhonghua liu xing bing...

Documents